Learn how to manage the files, symbols, and codebase resources the AI sees.
For an AI model to write accurate code, it requires relevant context. CoderX automatically handles context window allocation using background indexing and AST references. However, you can manually direct the model's attention using advanced context selectors.
When writing prompts in Workspace Chat (Ctrl+L) or Inline Edit (Ctrl+K), type @ to open the context dropdown:
Directly pins a specific file to the prompt context. Inserts the file's content along with line annotations.
Attaches all files within a directory, useful when explaining module patterns or component structures.
Pins a specific Class, function, interface, or struct. Leverages AST indexing to extract the symbol's definition.
Ingests git history, staged changes, unstaged changes, or specific commit logs for analysis.
When you open a workspace, CoderX builds a localized codebase map to support context searching:
The indexer parses your files to map symbol relationships, exports, definitions, and class structures.
Your code files are chunked and converted into semantic vector embeddings stored in a local sqlite database.
When you search or ask a question, CoderX runs a local cosine-similarity query on the database to locate the most relevant files instantly.
If you are working on a specific feature, you can pin assets in the Context Panel:
https://react.dev) to let the agent consult API standards on the fly..m0xignore)To keep your index small and save context space, create a .m0xignore file in your workspace root. Files matching these patterns will be skipped during indexing and codebase searches:
# Ignore build and dependency folders
node_modules/
.next/
dist/
# Ignore binary files and media
*.png
*.zip
*.tar.gz
# Ignore local keys
*.pem
secrets/